The present invention relates to a storage subsystem and a technique of controlling a storage subsystem, and in particular, to a technique of collecting status information etc., of a storage subsystem that is placed, for example, in a remote place and not directly coupled to a host computer.
As information-oriented society develops, computer systems are introduced all over the world, and data quantity processed there is increasing explosively. Further, importance of data treated in computer systems rises, and high reliability is required for keeping data, and it becomes social duty of an information system to prevent loss of data held in it by any kind of disaster.
In such a system, to ensure reliability of data, a storage itself is multiplied. Data outputted from a processing host computer is not only stored into a storage directly coupled to the host computer, but also copied to another storage via the directly coupled storage.
In the following, a storage directly coupled to a host computer is called a direct-coupled storage, and a storage that receives data not directly from the host computer but through the direct-coupled storage is called a remote storage.
Generally, a method of copying data to a remote storage through a direct-coupled storage is called remote copy, and applied to an information system that requires high reliability. According to this remote copy technique, even when a failure occurs in a storage so that the storage falls into an inoperable state, system operation can be continued using data of another storage.
When especially high reliability is required, it is possible to employ a method in which a multitude of remote storages are concatenated, and data is copied sequentially to the concatenated remote storages to increase multiplicity (redundancy) of data and to increase reliability.
For example, in the case of an apparatus for cascading and storing data, which is disclosed in Published Japanese Translation No. 2002-542526 of International Application (Patent Document 1), data is processed in a production site with a host computer and written into a direct-coupled storage and then, redundant data store is transparently provided to two or more remote storage. When a first remote storage receives data from the production site, the first remote storage transfers the data to a second remote storage located in a remoter place. At that time, the first remote storage holds a copy progress status. When data copy operation to the second remote storage is completed, the first remote storage reports the completion to the direct-coupled storage in the production site.
Further, to enhance integrity against disaster such as an earthquake, it is favorable that remote storages are located at remoter places as far away as possible. However, remote copy to a remote place takes time in data transfer. To solve this problem, there is a method in which, at the time of data update processing from a host computer, data transfer to remote storages is performed asynchronously with transfer (host transfer) from the host computer to a direct-coupled storage, to realize efficient remote copy between a plurality of storage subsystems, as disclosed, for example, in Japanese Patent Application Laid-Open No. 2002-334049 (Patent Document 2).
According to the technique disclosed in Patent Document 2, at a point of time when data transferred from the host computer is stored in an internal buffer of the direct-coupled storage, completion of reception is acknowledged to the host computer. And, thereafter, the host computer monitors the utilization factor of the internal buffer of the direct-coupled storage to adjust data update intervals.
Further, in the case where data is inherited at a point of time when a failure occurs in a storage, it is necessary to ensure consistency of data between multiplied storages. As a technique of establishing consistency of data, there is known a technique in which consistency of data contents is ensured as follows. Namely, a pair of volumes consisting of a volume (which stores data) of a direct-coupled storage and a volume (as the copy destination of the data mentioned) of a remote storage is defined, and a set of pairs of volumes are managed generally as a group, in order to maintain the order of data update, as disclosed, for example, in Japanese Patent Application Laid-Open No. 2002-189570.
The technique disclosed in Patent Document 3 is a duplication method to recover a volume copy quickly within the group when an event that temporarily stops copying, owing to maintenance or disaster. According to the present technique, a data center in a remote place can take over processing at the time when an information system suffers from disaster.